BigAlpha 2026 Final Scoring: Leaderboard and Jury Evaluation
Summary
The document describes how the BigAlpha 2026 final evaluates projects across factor discovery, end-to-end large models, and AI open innovation. A team's result combines a leaderboard-based component worth 30 points with a live judging component worth 70. The leaderboard component maps qualifying ranks to preset scores. Judges assess practical value, novelty, completeness, and presentation, applying evidence appropriate to each track, such as factor effectiveness or model prediction ability.
Judges score independently, with the highest and lowest total scores for each team removed before averaging the rest; judges with a conflict must recuse themselves. Final rankings span all three tracks within each event, with validity and robustness, research rigor, and leaderboard score used to break ties before a secret ballot. The document also specifies award adjustments for small finals and possible disqualification for serious violations. It explains an evaluation process rather than a trading method, and gives no evidence about the predictive or financial merit of any submitted work.
Key ideas
- Final scores combine a leaderboard component with a larger live judging component.
- Leaderboard points depend on a team's rank in its original track.
- Judges assess practical value, novelty, completeness, and presentation using track-specific evidence.
- The highest and lowest judge totals are removed before the remaining scores are averaged.
- Tie breaks prioritize result validity and robustness, then research rigor and leaderboard score.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.